from calibre.web.feeds.news import BasicNewsRecipe class NTVMSNBC(BasicNewsRecipe): title = u'NTV' __author__ = 'A Erdogan' description = 'News from Turkey' publisher = 'NTV' category = 'news, politics, Turkey' oldest_article = 7 max_articles_per_feed = 100 no_stylesheets = True use_embedded_content = False masthead_url = 'http://www.ntvmsnbc.com/images/MSNBC/msnbc_ban.gif' language = 'tr' remove_tags_before = dict(name='h1') remove_tags_after = dict(attrs={'id': 'haberDetayYazi'}) extra_css = ''' body{font-family:Arial,Helvetica,sans-serif; font-size:small; align:left; color:#000000} h1{font-size:large; color:#000000} h2{font-size:small; color:#000000} p{font-size:small; color:#000000} ''' conversion_options = { 'comment': description, 'tags': category, 'publisher': publisher, 'language': language } remove_tags = [dict(name=['embed', 'il', 'ul', 'iframe', 'object', 'link', 'base']), dict(name='div', attrs={'style': ['padding: 0pt 10px 10px;']}), dict(name='div', attrs={'style': ['padding: 0pt 10px 10px;']}), dict(name='div', attrs={ # noqa 'class': ['textSmallGrey w320']}), dict(name='div', attrs={'style': ['font-family:Arial; font-size:16px;font-weight:bold; font-color:#003366; margin-bottom:20px; margin-top:20px; border-bottom:solid 1px;border-color: #CCC; padding-bottom:2px;']})] # noqa remove_tags_before = dict(name='h1') remove_tags_after = dict(name='div', attrs={'style': [ 'font-family:Arial; font-size:16px;font-weight:bold; font-color:#003366; margin-bottom:20px; margin-top:20px; border-bottom:solid 1px;border-color: #CCC; padding-bottom:2px;']}) # noqa feeds = [(u'NTV', u'http://www.ntvmsnbc.com/id/3032091/device/rss/rss.xml')] def print_version(self, url): articleid = url.rpartition('/id/')[2] return 'http://www.ntvmsnbc.com/id/' + articleid + '/print/1/displaymode/1098/' def preprocess_html(self, soup): return self.adeify_images(soup)